Blind channel estimation for multi-user CDMA systems

Blind identification of digital communication channels has attracted a significant amount of attention. In this work, subspace-based methods are considered specifically for direct-sequence/code-division multiple-access (DS/CDMA) signals in an asynchronous, multipath, multi-user environment. The objective is to reduce the complexity of previously proposed methods by exploiting the properties of DS/CDMA waveforms. In particular, knowledge of the DS/CDMA spreading codes is employed, via matched filtering, to reduce the dimensionality. The resulting algorithm shows negligible difference in performance with respect to a previously proposed algorithm by Torlak and Xu (see IEEE Trans. on Signal Processing, vol.45, no.1, p.137-47, 1997). However, the reduction in complexity can be significant. The Cramer-Rao lower bounds on the estimation variance are determined and provide further justification for the matched filtering scheme